Fredrik Höglund <fred...@kde.org> writes: > This extension decouples the vertex attributes from the vertex buffer > bindings, making it possible to change buffers without having to > respecify the formats. The mapping between attributes and vertex > buffer bindings also becomes configurable.
Patch 5 as posted doesn't compile. I applied this:
@@ -105,16 +105,18 @@ type_to_bit(const struct gl_context *ctx, GLenum type)
static inline struct gl_vertex_attrib_array *
gl_vertex_attrib_array(struct gl_context *ctx, GLuint index)
{
+ struct gl_array_object *arrayObj = ctx->Array.ArrayObj;
ASSERT(index < Elements(arrayObj->VertexAttrib));
- return &ctx->Array.ArrayObj->VertexAttrib[index];
+ return &arrayObj->VertexAttrib[index];
}
Also, when posting a big series that's likely to conflict soon, it's
nice to put it in a personal repo so people can just grab it instead of
applying it off the list. I had to resolve a few conflicts from atomic
counters.
> This series adds new state structs and changes the gl*Pointer()
> functions to update the new states as described in the specification.
> To avoid making invasive changes in the vbo module I decided to keep
> the gl_client_arrays for now. Those are now derived from the
> gl_vertex_attrib_array and gl_vertex_buffer_binding states.
> In theory this should also make it possible to incrementally port
> other parts of Mesa to use the new states directly.
I decided to check if the overhead of the computed state was a problem,
and on INTEL_NO_HW=1 cairo-gl I saw no effect (n=3).
> Note that this adds a glVertexAttribLFormat() entry point.
> Mesa doesn't support double-precision attributes (yet), but the
> GL specification says that the results of using a glVertex*L*()
> function to set a non-double precision attribute are undefined,
> so I don't think this is a problem.
>
> The extension is written against the GL 4.2 specification, and
> based on the tables it refers to one can infer that it requires
> ARB_vertex_array_bgra, ARB_vertex_type_2_10_10_10_rev and
> ARB_ES2_compatibility. I think the dependencies section should
> have language that removes references to vertex formats added by
> those extensions when they're not supported, so in this series
> I have enabled the extension unconditionally.
Agreed.
pgpeeoEwRvqzP.pgp
Description: PGP signature
_______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
